diff mbox

[2/2] generic: test I/O on dm error device

Message ID 1457710822-30532-2-git-send-email-eguan@redhat.com
State Not Applicable
Headers show

Commit Message

Eryu Guan March 11, 2016, 3:40 p.m. UTC 
This is a test that performs simple I/O on dm error device, which
returns EIO on all I/O request.

This is motivated by an ext4 bug that crashes kernel on error path when
trying to update atime. Following kernel patch should fix the issue

  ext4: fix NULL pointer dereference in ext4_mark_inode_dirty()

Signed-off-by: Eryu Guan <eguan@redhat.com>
---
 tests/generic/338     | 90 +++++++++++++++++++++++++++++++++++++++++++++++++++
 tests/generic/338.out |  2 ++
 tests/generic/group   |  1 +
 3 files changed, 93 insertions(+)
 create mode 100755 tests/generic/338
 create mode 100644 tests/generic/338.out

Comments

Dave Chinner March 15, 2016, 2:46 a.m. UTC | #1 
On Fri, Mar 11, 2016 at 11:40:22PM +0800, Eryu Guan wrote:
> This is a test that performs simple I/O on dm error device, which
> returns EIO on all I/O request.
> 
> This is motivated by an ext4 bug that crashes kernel on error path when
> trying to update atime. Following kernel patch should fix the issue
> 
>   ext4: fix NULL pointer dereference in ext4_mark_inode_dirty()

Why does this test require the loop device? Why can't you just
unmount the filesystem, run 'blkdev --flushbufs <dev>' to ensure
there are no cached buffers/pages on the block device, then mount
it again?

Cheers,

Dave.
Eryu Guan March 15, 2016, 8:02 a.m. UTC | #2 
On Tue, Mar 15, 2016 at 01:46:16PM +1100, Dave Chinner wrote:
> On Fri, Mar 11, 2016 at 11:40:22PM +0800, Eryu Guan wrote:
> > This is a test that performs simple I/O on dm error device, which
> > returns EIO on all I/O request.
> > 
> > This is motivated by an ext4 bug that crashes kernel on error path when
> > trying to update atime. Following kernel patch should fix the issue
> > 
> >   ext4: fix NULL pointer dereference in ext4_mark_inode_dirty()
> 
> Why does this test require the loop device? Why can't you just
> unmount the filesystem, run 'blkdev --flushbufs <dev>' to ensure
> there are no cached buffers/pages on the block device, then mount
> it again?

Yes, 'blockdev --flushbufs <dev>' works, and I found that I only need to
add a blockdev call before dropping caches. This makes the code much
cleaner and easier to read, perhaps the first patch can be dropped as
well. I'll send out v2 shortly. Thanks for the review!

Eryu
--
To unsubscribe from this list: send the line "unsubscribe linux-ext4" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html
diff mbox

Patch

diff --git a/tests/generic/338 b/tests/generic/338
new file mode 100755
index 0000000..cea4d82
--- /dev/null
+++ b/tests/generic/338
@@ -0,0 +1,90 @@ 
+#! /bin/bash
+# FS QA Test 338
+#
+# Test I/O on dm error device.
+#
+# Motivated by an ext4 bug that crashes kernel on error path when trying to
+# update atime.
+#
+#-----------------------------------------------------------------------
+# Copyright (c) 2016 Red Hat Inc.,  All Rights Reserved.
+#
+# This program is free software; you can redistribute it and/or
+# modify it under the terms of the GNU General Public License as
+# published by the Free Software Foundation.
+#
+# This program is distributed in the hope that it would be useful,
+# but WITHOUT ANY WARRANTY; without even the implied warranty of
+# MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+# GNU General Public License for more details.
+#
+# You should have received a copy of the GNU General Public License
+# along with this program; if not, write the Free Software Foundation,
+# Inc.,  51 Franklin St, Fifth Floor, Boston, MA  02110-1301  USA
+#-----------------------------------------------------------------------
+#
+
+seq=`basename $0`
+seqres=$RESULT_DIR/$seq
+echo "QA output created by $seq"
+
+here=`pwd`
+tmp=/tmp/$$
+status=1	# failure is the default!
+trap "_cleanup; exit \$status" 0 1 2 3 15
+
+_cleanup()
+{
+	cd /
+	rm -f $tmp.*
+	_dmerror_cleanup
+	_destroy_loop_device $LOOP_DEV
+	rm -f $LOOP_FILE
+	rm -rf $LOOP_MNT
+}
+
+# get standard environment, filters and checks
+. ./common/rc
+. ./common/filter
+. ./common/dmerror
+
+# remove previous $seqres.full before test
+rm -f $seqres.full
+
+# real QA test starts here
+_supported_fs generic
+_supported_os Linux
+_require_test
+_require_loop
+_require_dm_target error
+# If TEST_DEV is not a valid block device, FSTYP cannot be mkfs'ed either
+_require_block_device $TEST_DEV
+
+echo "Silence is golden"
+
+# Use loop device as backend of dm error device, because drop_caches drops
+# caches hold by loop device too and forces reading inode info from disk and
+# triggers NULL pointer dereference on buggy ext4
+LOOP_FILE=$TEST_DIR/$seq-$$.img
+LOOP_MNT=$TEST_DIR/$seq-$$.mnt
+mkdir -p $LOOP_MNT
+$XFS_IO_PROG -fc "truncate 512M" $LOOP_FILE >>$seqres.full 2>&1
+LOOP_DEV=`_create_loop_device $LOOP_FILE`
+
+_dmerror_init $LOOP_DEV $LOOP_MNT
+_mkfs_dev $DMERROR_DEV
+# Use strictatime mount option here to force atime updates, which could help
+# trigger the NULL pointer dereference on ext4 more easily
+_dmerror_mount "-o strictatime"
+_dmerror_load_error_table
+
+# drop all caches, force reading from error device
+echo 3 > /proc/sys/vm/drop_caches
+
+# do some test I/O
+ls -l $LOOP_MNT >>$seqres.full 2>&1
+$XFS_IO_PROG -fc "pwrite 0 1M" $LOOP_MNT/testfile >>$seqres.full 2>&1
+
+# no panic no hang, success, all done
+status=0
+exit
diff --git a/tests/generic/338.out b/tests/generic/338.out
new file mode 100644
index 0000000..3482cf4
--- /dev/null
+++ b/tests/generic/338.out
@@ -0,0 +1,2 @@ 
+QA output created by 338
+Silence is golden
diff --git a/tests/generic/group b/tests/generic/group
index 727648c..8818827 100644
--- a/tests/generic/group
+++ b/tests/generic/group
@@ -340,3 +340,4 @@ 
 335 auto quick metadata
 336 auto quick metadata
 337 auto quick metadata
+338 auto quick rw